feat: improve todo comment identification

src/identify_todo_comment.rs

@@ -1,11 +1,45 @@
-use crate::todo_keywords::TODO_KEYWORDS;
+pub const PRIMARY_TODO_KEYWORDS: [&str; 9] = [
+    "TODO", "FIXME", "CHANGED", "XXX", "HACK", "BUG", "OPTIMIZE", "REFACTOR",
+    "TEMP",
+];
+
+pub const SECONDARY_TODO_KEYWORDS: [&str; 13] = [
+    "IDEA",
+    "NOTE",
+    "REVIEW",
+    "NB",
+    "QUESTION",
+    "DEBUG",
+    "KLUDGE",
+    "COMPAT",
+    "WARNING",
+    "DANGER",
+    "INFO",
+    "DEPRECATED",
+    "COMBAK",
+];
 
 pub fn identify_todo_comment(comment_text: &str) -> Option<String> {
-    let trimmed_text = comment_text.trim().to_uppercase();
-    for keyword in TODO_KEYWORDS.iter() {
-        if trimmed_text.starts_with(keyword) {
-            return Some(keyword.to_string());
+    let trimmed_text = comment_text.trim();
+
+    let words_with_separators: Vec<&str> =
+        trimmed_text.split_whitespace().collect();
+
+    for keyword in PRIMARY_TODO_KEYWORDS.iter() {
+        for word in &words_with_separators {
+            if word.to_uppercase().contains(&keyword.to_uppercase()) {
+                return Some(keyword.to_string());
+            }
         }
     }
+
+    if let Some(first_word) = words_with_separators.first() {
+        for keyword in SECONDARY_TODO_KEYWORDS.iter() {
+            if first_word.to_uppercase() == keyword.to_uppercase() {
+                return Some(keyword.to_string());
+            }
+        }
+    }
+
     None
 }

src/lib.rs

@@ -16,5 +16,4 @@ pub mod identify_supported_file;
 pub mod identify_todo_comment;
 pub mod prepare_blame_data;
 pub mod remove_duplicate_dates;
-pub mod todo_keywords;
 pub mod types;

tests/identify_todo_comment.rs

@@ -0,0 +1,73 @@
+use todoctor::identify_todo_comment::identify_todo_comment;
+
+#[tokio::test]
+async fn test_primary_keyword_found() {
+    let comment = "// TODO: This is a test comment.";
+    assert_eq!(identify_todo_comment(comment), Some("TODO".to_string()));
+}
+
+#[tokio::test]
+async fn test_primary_keyword_middle_of_sentence() {
+    let comment = "// This is a test TODO comment.";
+    assert_eq!(identify_todo_comment(comment), Some("TODO".to_string()));
+}
+
+#[tokio::test]
+async fn test_primary_keyword_at_end() {
+    let comment = "// This is a test TODO";
+    assert_eq!(identify_todo_comment(comment), Some("TODO".to_string()));
+}
+
+#[tokio::test]
+async fn test_primary_keyword_lowercase() {
+    let comment = "// todo: lowercase todo comment.";
+    assert_eq!(identify_todo_comment(comment), Some("TODO".to_string()));
+}
+
+#[tokio::test]
+async fn test_secondary_keyword_at_start() {
+    let comment = "// FIXME: There is a bug.";
+    assert_eq!(identify_todo_comment(comment), Some("FIXME".to_string()));
+}
+
+#[tokio::test]
+async fn test_secondary_keyword_not_found_in_middle() {
+    let comment = "// This is a REVIEW comment.";
+    assert_eq!(identify_todo_comment(comment), None);
+}
+
+#[tokio::test]
+async fn test_comment_without_keywords() {
+    let comment = "// Just a regular comment.";
+    assert_eq!(identify_todo_comment(comment), None);
+}
+
+#[tokio::test]
+async fn test_primary_keyword_with_punctuation() {
+    let comment = "// bla bla -- TODO: fix this.";
+    assert_eq!(identify_todo_comment(comment), Some("TODO".to_string()));
+}
+
+#[tokio::test]
+async fn test_secondary_keyword_case_insensitive() {
+    let comment = "// fixme: case insensitive.";
+    assert_eq!(identify_todo_comment(comment), Some("FIXME".to_string()));
+}
+
+#[tokio::test]
+async fn test_primary_keyword_with_dot() {
+    let comment = "// This comment has a todo in the middle.";
+    assert_eq!(identify_todo_comment(comment), Some("TODO".to_string()));
+}
+
+#[tokio::test]
+async fn test_secondary_keyword_with_case_variation() {
+    let comment = "// FiXmE: Mixed case keyword at the start.";
+    assert_eq!(identify_todo_comment(comment), Some("FIXME".to_string()));
+}
+
+#[tokio::test]
+async fn test_no_keyword_found() {
+    let comment = "// This is just a comment without anything.";
+    assert_eq!(identify_todo_comment(comment), None);
+}